Parity Check 2456 Errors - MemTest ok - Parity Check with corrections does not Help - Whats Next?

[stekay]

Hey guys, 

 

I have a problem with my unraid server. The Parity Check gives me 2456 Errors. 

What I can tell: 

 

It started at the 20.07.2021.

At the 19.07.2021 I had a warning (I have the history in my Telegram Bot Message...):

---------

SKUNRAID: Notice [SKUNRAID] - array health report [PASS] Array has 5 disks (including parity & cache)

00:20

SKUNRAID: Warning [SKUNRAID] - array has errors Array has 1 disk with read errors

01:29

SKUNRAID: Notice [SKUNRAID] - current pending sector returned to normal value WDC_WD20EARX-00PASB0_WD-WMAZA5358183 (sdd)

01:46

SKUNRAID: Notice [SKUNRAID] - Parity check finished (0 errors) Duration: 8 hours, 51 minutes, 41 seconds. Average speed: 94.1 MB/s

06:05

July 20

---------

20.07.2021

 

SKUNRAID: Notice [SKUNRAID] - array turned good Array has 0 disks with read errors

09:10

SKUNRAID: Notice [SKUNRAID] - Parity check started Size: 3 TB

09:10

SKUNRAID: Notice [SKUNRAID] - Parity check finished (2456 errors) Duration: 8 hours, 51 minutes, 24 seconds. Average speed: 94.1 MB/s

18:02

 

 

So my question is: What can I do now?

I did a Memtest with no errors. I restarted the server, I did a Parity check with Corrections enabled. (Btw. I saw, that this was also enabled in the monthly, sheduled check...I Unchecked it there for now)

 

I forgot to mention: We had a couple of power shutdowns here... so My server... was shut down a couple of times...maybe this is the reason I have problems now? And I think we also had one back in July at when those errors popped up.

 

Thank you in advance!

 

 

 

 

Edited August 17, 2021 by stekay

[JorgeB]

If you haven't rebooted since last parity check please post diags: Tools -> Diagnostics

[stekay]

5 minutes ago, JorgeB said:

If you haven't rebooted since last parity check please post diags: Tools -> Diagnostics

Nope. I did a reboot after the check last night to make the memtest. So should I start a parity check now again without correction and post a log when this is finished ?

[JorgeB]

9 minutes ago, stekay said:

So should I start a parity check now again without correction and post a log when this is finished ?

Yep, but not just the log, complete diags.

[JorgeB]

Also note that if you only did one correcting check it's normal it would find the same errors, but next one should find 0.

[stekay]

1 minute ago, JorgeB said:

Also note that if you only did one correcting check it's normal it would find the same errors, but next one should find 0.

Thanks. I will do it. 
 

like I mentioned, in the sheduled parity check there was correction enabled. So the errors would have been gone on the last check last night, wouldn’t them ?

[JorgeB]

If that's correct yes.

[stekay]

8 hours ago, JorgeB said:

If that's correct yes.

Okay, the Parity Check is done - 0 Errors. Now I´m very confused.

I mean, why are there 0 Errors now - sure - I manually startet it. But like you can see, there are def. 2456 Errors 4 times - and in the sheduled check, correction was def. enabled.

Don´t understand me wrong, I´m happy that it looks like there is no Error, but I would like to understand, why this happened. Is the "Standard-Setting" in the Sheduled Parity-Check buggy?

 

Thank you for your help!

 

skunraid-diagnostics-20210817-2123.zip

 

[itimpi]

When do you have the scheduled check set to run?    I would guess from the dates they were more likely unscheduled ones?   If they were ones run after an unclean shutdown then that defaults to non-correcting.

[stekay]

25 minutes ago, itimpi said:

When do you have the scheduled check set to run?    I would guess from the dates they were more likely unscheduled ones?   If they were ones run after an unclean shutdown then that defaults to non-correcting.

Good point, I see what you mean. The sheduled ones are set to run at every first day in the month. So...those dates, where those errors occurred are those days when the system was shut down because of my power shutdowns....So Unraid is doing a parity check after an unclean parity check on its own? And those checks are without correction? Then everything would be clear and it would explain, why the errors are gone now: Because I started a check WITH correction last night. And one without today.

 

Is it better to set those sheduled checks to not correct? Or should I activate it again for the sheduled?

 

Best and thank you very much

[itimpi]

UnRaid automatically runs a non-correcting parity check the next time you boot after an unclean shutdown.   If you are regularity getting unclean shutdowns then you need to investigate why, and perhaps invest in a UPS if they are due to power interruptions.

 

it is recommended that scheduled checks are non-correcting.    This is because if a drive is playing up and returning bad data you might end up corrupting parity.

[stekay]

1 minute ago, itimpi said:

UnRaid automatically runs a non-correcting parity check the next time you boot after an unclean shutdown.   If you are regularity getting unclean shutdowns then you need to investigate why, and perhaps invest in a UPS if they are due to power interruptions.

 

it is recommended that scheduled checks are non-correcting.    This is because if a drive is playing up and returning bad data you might end up corrupting parity.

Thanks for the explenation!

 

Alright. Yeah, normally we don´t have any problems here, it was because of the high water...but I will def. think about a UPS.

Alright, I let the check correction deactivated and just correct manually then.

Is there a topic somewhere, where I can read into it a bit? So for example more about stuff like if a drive is playing up...etc... ?

 

Thanks for your help! Very appreciate it!

[trurl]

57 minutes ago, stekay said:

So for example more about stuff like if a drive is playing up...etc... ?

If the Errors column in Main isn't zero for one of your data disks, for example, then that disk could be the reason parity check finds sync errors, so you don't want to correct parity based on bad data from that other disk.

 

Or SMART warnings on the Dashboard for a disk, for another example.

 

Further investigation would involve looking at SMART for the disk, checking syslog to try to determine if the cause is bad connection (most often) or something else.

 

Or just posting your diagnostics for us to take a look.

[stekay]

8 hours ago, trurl said:

If the Errors column in Main isn't zero for one of your data disks, for example, then that disk could be the reason parity check finds sync errors, so you don't want to correct parity based on bad data from that other disk.

 

Or SMART warnings on the Dashboard for a disk, for another example.

 

Further investigation would involve looking at SMART for the disk, checking syslog to try to determine if the cause is bad connection (most often) or something else.

 

Or just posting your diagnostics for us to take a look.

Thanks a lot